Heated Ferrite == Useless Ferrite

[GelatinCapacitor]

Thanks YouTube videos (mainly diodegonewild), I was convinced that modifying a ferrite transformer was easy. Just heat up the core to destroy the glue that holds the core together.

However, I have had many repair attempts where I had to separate the core of small power supplies to modify the windings. All of them have failed. And I always put the blame on something besides the transformer.

Not long ago, I began using a small flyback SMPS with its transformer taken away to test other flyback transformers. A simple idea. And that let me notice that taking apart the transformer killed it every time.

So, after that I performed an experiment. I sacrificed many flyback SMPS to unglue their cores through various methods (boiling, air, iron), and then test them again with to see which of them survive.

None of them survive.

So, simply heating the core until it can be separated, then putting it back together, without modifying anything else, makes the transformer useless. Even with the iron method, in which anything other than the core barely gets hot.
I did my test with small flyback SMPSs, in the range of a tiny 5W USB charger to a 90W CRT TV power supply.

I found a question on StackExchange which declares the same conclusion.
https://electronics.stackexchange.com/questions/497489/temperature-degradation-of-ferrite-cores

I had originally planned to modify an ATX supply to suit my needs by modifying the transformer this way. That's cancelled now.
One of the comments in that StackExchange question says that bigger power transformers don't suffer from this problem, but I'm not willing to test that out.

My iron method consists of making the core make contact with the soldering iron, and over the span of ~20 min., slowly raise the temperature to ~160°C.
The core may be ready to take apart way before that. The signal I take is a certain smell, which I assume is the glue giving up.

After I discovered all of this, I have made many transformer mods without taking the core apart, even though painfully and slowly, and all of them have been successful.


Customizing windings without opening the core

[amyk]

Even that link you mentioned shows that ferrites are not affected by the sort of temperatures you're using. My guess is that you are not reassembling them correctly. Even a tiny air gap will have a huge effect.

[Kleinstein]

Soft ferrites are not permanently effected by heat (unless you go to the extremes near 1000 C). They however have a relatively low curie temperature and thus don't work as long as they are hot (e.g. > 150 C). This effect is however reversible. The air gap can be a big issue - there would have been a spacer that is not much different from the glue.

In therory there could be flybacl transformers with a permanent magnet premagnetization, that could be effected by heat. However I have not seen such a thing, except for linearity correction with CRTs. So I doubt one would find it in more normal PC supply - though they would do all kind of tricks to save a few cents.

[Berni]

Airgaps are very important in ferrite transformers.

The ferrite saturates much faster than classical iron lamination while having a high permeability. So airgaps have a larger effect on the behavior of a transformer.

Transformers designed for forwards converters or gate drive and similar typically have no air gap as they are not meant to store energy, just pass it on instantly. So for these transformers it is important to assemble them as tightly as possible to avoid creating an unnecessary airgap, otherwise the transformer might not reach the designed inductance.

Transformers designed for flyback converters and similar do tend to have a precisely designed intentional airgap. They are designed for storing energy between switching cycles, but ferrite is fairly bad at this, so they introduce an airgap to provide a place to store the energy. While this does reduce the inductance, it also increases the saturation current by a lot. The size of this airgap is carefully designed to make a good compromise of both. If you remove the carefully designed in air gap then the transformer might just saturate when used in a flyback topology.

[Ian.M]

To avoid confusion, the 'airgap' doesn't usually contain air! Except for cores gapped by grinding the center limb, its more usual for the gap to be filled with an insulator of some sort to provide the required spacing.

Additionally, the core may be gapped by mixing a small percentage of tiny glass spheres into the glue, with the size of the spheres chosen for the required gap.   It can be extremely difficult to distinguish between a core gapped this way, and an ordinary glue line.   

Unless you measure the outside dimension of the core, across the glue line or gap, with a micrometer (after removing tape etc. and cleaning the exterior surfaces) before disassembly, and measure again after reassembly, you wont know how accurately you have re-established the original core gap (or lack thereof).

[mag_therm]

Refer to Fair-Rite "Product Life"
When re-using salvaged cores, it might be necessary to know at least whether it was a MnZn ( low freq, power transformer etc), or Ni Zn (high Freq)
Both types can be permanently damaged by excess magnetic field strength H swing   and mechanical stress , NiZn may be more susceptible to those.
Both types should recover after being heated above Curie Temperature.
MnZn Cores usually work most efficiently  at around 90 C There are graphs for the material types showing this. So design for 70 to 80 C helps counteract thermal runaway effects.

For hobby here I have hand wound about 20 power transformers, and  inductors using surplus  (new old stock) MnZn cores from electronics surplus suppliers.
For components in the 10's MHz range needing NiZn I would always buy new cores. I have had good success just using basic design methods although some I have done with FEM.
I hav e FairRite Type 61 toroids running at 14MHz as power transformers for ham radio. They run too hot to touch, and they are efficient enough. It is a  bit difficult to measure temperature and performance in situ  at that frequency.

If a new core set is around $2 I wonder why salvage attempts  would be worthwhile.

[strawberry]

flyback winding polarity is crucial and proper airgap too
it will lead to blown or overheating switching device (not loaded)
and wrong turns and incorrect layer orders will lead to poor performance or safety (EMI, less power/voltage..)

applying heat ~160deg C may damage winding insulation(cheap wire is rated max op 150)

[GelatinCapacitor]

I definitely put effort into making sure the cores were put together the way they used to, and in the best way I could, and to hold them tightly with kapton tape.

For most of my attempts, I separated and put the core together many times just in case I did something wrong.

All of the cores kept glue were it used to be, including the center gap. I did not try to take away any glue. All of the attempts were made once everything completely cooled down. At least visually, all of the cores seemed to fit back together perfectly, with no additional gaps or anything. Perfect lego pieces.

Many of the cores were not even perfectly aligned to begin with. I replicated that misalignment just in case. I also tried better alignments for the same cores.

A couple of the cores had little square or circular shaped spacers in the middle. Of course I did not remove those. All of these cores were difficult NOT to align correctly.

For a few of the cores, I did not even take the cores apart, just made sure I got the 'ready to take apart' signal. Same results.

For all transformers, after declaring them dead, I removed all windings and, not very properly tested for loss of enamel. I did not detect any evidence of loss of enamel, but maybe the way I tested this is completely off. In the iron tests, I did not measure temperatures above 120°C on the cores and 60°C on windings, so I do not suspect the windings in the iron tests.

If I were to redo the tests, I would now also try to measure the resistance of the windings to check for loop shorts. By running current across them before and after heating.

Maybe even take the windings away before heating, and then put them back afterwards. That would drive me nuts.

[strawberry]

shorted one of windings = will lose all inductance except small parasitic inductance(leakage)
if you dont have LCR meter use functiongen and oscilloscope and observe transformer waveform
if core is broken somehow = low inductance (inductance equal that of air/no core)

[Buriedcode]

Regarding re-using old ferrites and disassembling them: I've had some success with solvents. Namely paint thinner/white spirit.  The silicone/rubbery stuff is unaffected, but varnish and tape residue will dissolve in 24 hours.  As for heat, I've used a hot air gun - the larger 2kW kind, on a low setting, far enough away so it doesn't melt the bobbin - but its very difficult to get it right, plus its a proper hassle handling the hot cores, so I quickly gave up that method.

The solvent has worked for all ferrites salvaged from ATX supplies, but more modern SMPS transformers - specifically the sammler ones - took a lot longer.

I got a joblot of small EFD20 ferrites off ebay years ago, and every so often I'll "pickle" a batch to separate them.  I only have a few projects using those, but they worked as expected (had to change the airgap on a few, or completely remove the gap entirely), although I had to measure the core parameters to make sure they were in spec.

[GelatinCapacitor]

It would be cool if all transformers were held together by easy to remove parts.
But glue solves many problems at once.


I always got rid of these, even if I got their details (from SONY CRTs):

[moffy]

An interesting discussion, but I haven't seen @GelatinCapacitor how the cores have failed. Do they have reduced permeability or saturation? Some quantitive measurements would be of interest and might point to a possible cause.

[GelatinCapacitor]

Thank you @moffy. I'm just a random person that has many science related hobbies. Sometimes I earn my money from performing electronic repairs, but I've still got no oscilloscope, func. gen., or LCR meter. Not because I don't want them, but because I wouldn't really use them for now.
Maybe it's time to get that dang LCR meter. Whenever I get it, I will absolutely remember to make a follow up.

For now I think the point is clear:
transformers are difficult to dissasemble

Also, I assume the cores lost permeability, because I test them with an SMPS that uses some tipycal IC: if I short ANY winding of a good transformer (considering that the primary on all of these tested transformers had an unused center tap, and the secondary was in between these primary halves; considering each primary half as a unique winding), I could at least measure some voltage pulses from other windings, from the IC's attempts to start. But with the after-heated cores, it's like the core isn't even there. 

Sorry for my rudimentary methods. 

[moffy]

Well, I look forward to any follow up info you might provide.

[T3sl4co1l]

This thread is useless without measurements

You don't need to use an SMPS (and definitely shouldn't) to do that testing.  A signal generator, capacitor, resistor and oscilloscope do a fine job.  A pulse tester can also be built, usually using a signal generator (or in a pinch, a 555 timer), MOSFET, diode, capacitors, shunt resistor, and power supply.  Example on my old website, https://www.seventransistorlabs.com/tmoranwms/Elec_Magnetics.html which, the magnetic theory part I think is still fine, if a bit out of style for me these days, and I do have better and more diverse inductor and transformer design methods today (but, also which aren't worth going into detail on such a page).

After fractures (which can be glued back together with superglue or epoxy, with care), probably burrs and varnish are the next most common fault.  Use a knife to scrape off the excess, and use a very flat sharpening stone or lap to grind the core faces flat.  If you find only the outer limbs are grinding, congratulations -- you have an air-gapped core, and the inductivity (A_L) will be low; this is suitable for flyback converters and inductors, and unsuitable for forward converters and CMCs.

Occasionally you'll see an E-core (or other shapes) that is definitely not air-gapped, and yet doesn't have a high A_L; I haven't seen these very often (and, I seem to have lost the one example that I had found..!), but, these shapes are occasionally made in powdered-iron and related materials, with lower permeability, so can be used ungapped just as any powder toroid.

As for leakage, or coupling coefficient, note this also depends upon the relative positioning of windings.  The core is not an ideal magnetic path (and even if it were, it wouldn't actually matter--), and there are plenty of flux paths outside of it, literally leaking around; wires hanging around anywhere in the space around the core, can pick up varying amounts of those paths.  You maximize coupling when the windings are, as much as possible, in the same place: adjacent layers, twisted pair or bifilar, alternating layers or strands in parallel, etc.

Tim

[moffy]

Example on my old website, https://www.seventransistorlabs.com/tmoranwms/Elec_Magnetics.html which, the magnetic theory part I think is still fine, if a bit out of style for me these days, and I do have better and more diverse inductor and transformer design methods today (but, also which aren't worth going into detail on such a page).

Thanks, saved the page as a reference, always nice to have something to assist with memory.

[wraper]

Soft ferrites are not permanently effected by heat (unless you go to the extremes near 1000 C). They however have a relatively low curie temperature and thus don't work as long as they are hot (e.g. > 150 C). This effect is however reversible. The air gap can be a big issue - there would have been a spacer that is not much different from the glue.

In therory there could be flybacl transformers with a permanent magnet premagnetization, that could be effected by heat. However I have not seen such a thing, except for linearity correction with CRTs. So I doubt one would find it in more normal PC supply - though they would do all kind of tricks to save a few cents.

It's certainly not as much as 1000^oC as I've seen cooked transformers with useless cores and no way they heated up that much while having bobbin remaining in one piece or FR-4 PCB not catching flames under toroidal one. But it's not 160^oC either. 

[strawberry]

ferrite is made by sintering